Road surface condition monitoring with an optical fibre reservoir structure based on acceleration data from vehicles

Abstract

The development of high-efficiency road pavement condition monitoring systems is critical for road asset management to obtain a more efficient evaluation of the health condition. Traditional specialized road assessment vehicles provide accurate road evaluations, but the limited vehicle number become a barrier to fast updating road surface condition data for road management decision-making. In this study, a physical optical computing approach for monitoring road surface conditions using only acceleration data from civilian vehicles was proposed. By processing the acceleration data with an optical fibre-based reservoir computing structure, road pavement surface fluctuation slopes and the international roughness indices were predicted. The predicted road surface condition results demonstrate the effectiveness of this model-free, physical reservoir computing approach in predicting road surface conditions. About 90% road surface fluctuations peaks and valleys detection were predicted. The accuracies of prediction of international roughness index for 100 m international roughness indices averaging and for international roughness indices (10 m) were about 86% and 73%. This method increases the efficiency of road condition monitoring and identifies potential critical road monitoring regions to lighten the road survey burden of specialized road assessment vehicles. This research focuses on expanding the application field of optical reservoir computing to the road monitoring field and provides primary research for the development of physical reservoirs for road surface monitoring on vehicles in the future.